Reproduce “From Sessions to Serving: Integrating LLM-Based Embeddings into Retrieval for Production Content Recommendation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhixiong Hu, Meng Li, Maggie Zhang, Deborshi Goswami, David Bailey, James Bradley
LSER augments Autodesk’s two-tower retrieval system with large language model embeddings of recent-session and longer-term user activity, while retaining learned sequence encoders. It refreshes these vectors asynchronously for serving. In an offline ablation, the added vectors raised popular-products HR@10 by 9.6 percentage points and MRR by 0.021; gains varied across surfaces, and the online test did not enable them.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.